The Case for Modularity: Sin or Salvation?

HAL Id: ijn_00000135

https://jeannicod.ccsd.cnrs.fr/ijn_00000135

Submitted on 4 Sep 2002

HAL

is a multi-disciplinary open access archive for the deposit and dissemination of sci- entific research documents, whether they are pub- lished or not. The documents may come from

L’archive ouverte pluridisciplinaire

destinée au dépôt et à la diffusion de documents scientifiques de niveau recherche, publiés ou non, émanant des établissements d’enseignement et de

The Case for Modularity: Sin or Salvation?

Scott Atran

To cite this version:

Scott Atran. The Case for Modularity: Sin or Salvation?. Evolution and Cognition, 2001, 7 (1),

pp.46-55. �ijn_00000135�

Introduction

The central thesis of PANKSEPP and PANKSEPP

(2001, henceforth, P&P) is that there are no spe- cially evolved, higher-or- der domain-specific cognitive mechanisms to speak of. Apart from the basic DARWINIAN emo- tions (fear, surprise, an- ger, sadness, joy, disgust), so-called cognitive ‘mod- ules’ are more likely “the product of dubious hu- man ambition rather than sound scientific rea- soning”. What evolution- ary psychologists (as well as many cognitive and developmental psycholo- gists) take to be modular structures are actually the epigenetic products of the ancestral emotional functions of the brain (rooted in what was once called “the limbic sys- tem,” MACLEAN 1990) and more recent “gen- eral-purpose brain mech-

anisms”. In more than a dozen places, P&P repeat as mantra that empirical evidence indicates the mod- ern human mind was created through the dual functioning of subcortical mammalian emotions and a neocortical general-purpose computational device. Furthermore, this highly flexible, all-pur- pose intelligence probably emerged “more rapidly via group selection than by individual selection”.

There is much to commend the cautionary tale that P&P tell regarding possible excesses of sociobi- ological speculation, including what are arguably

‘just-so’ stories about modularized adaptations for

rape (THORNHILL/PALMER

2000), homicide (BUSS

1999) and emotional dis- orders such as depression (NESSE/LLOYD 1992).

P&P’s position on the de- rivative and epigenetic character of ‘secondary’

or ‘social’ emotions (guilt, love, empathy, etc.), is also defensible (e.g., DAM- ASIO 1994), although seri- ously debatable (cf.

LEDOUX 1996; GRIFFITHS

1997). Finally, P&P’s take on the current, overly- funded fad for neuroim- aging (driven more by technological innovation and the industry it sup- ports than by any theoret- ical insight) is credible.

There can no more be a theory directly derived from observations of neu- roimaging than there can be a theory derived di- rectly from observations of meter readings; it is only in the service of some prior abstract the- ory that such observa- tions can make sense. Nevertheless, there is no em- pirical evidence whatever to indicate that P&P’s central thesis is true, and much to suggest it is false.

Evolutionary Psychology:

No Explanatory Value for Higher-Order Cognition (Yet)

To-date, evolutionary psychology has not predicted or discovered any significant or surprising aspect of language or higher-order cognitive devices of the sort I describe below. But this does not mean that all

Scott Atran

The Case for Modularity: Sin or Salvation?

The case for evolved cognitive modules rests on sever- al converging lines of evidence: Functional design (analogy), ethology (homology), universality, precoci- ty of acquisition, independence from perceptual expe- rience (poverty of stimulus), selective pathology (cerebral impairment), resistance to inhibition (hyper- activity), ease of cultural transmission. No factor may be necessary but evidence for all or some is compel- ling, if not conclusive, for domains like folkmechan- ics, folkbiology, folkpsychology. By constrast, no empirical evidence supports PANKSEPP and PANKSEPP’s central thesis: that what evolutionary psychologists (and many cognitive and developmental psycholo- gists) consider modular structures are actually epige- netic products of subcortical mammalian emotions and neocortical general-purpose computations. Argu- ably, no significant empirical discovery about lan- guage or other higher-order cognitive structures yet owes to inquiries about evolutionary origins and func- tions. Nevertheless, adopting evolutionary psycholo- gy’s requirement that candidate exaptations and spandrels be described, as far as possible, in connec- tion with evolved adaptations, opens new avenues for exploring and testing modular designs.

Modularity, domain-specificity, language, folkpsy- chology, folkbiology.

Abstract

Key words

Scott Atran

evolutionary psychology accounts are ‘just-so’ sto- ries. The best account of language as an evolution- ary adaptation centers on the claim that “language shows signs of design for the communication of propositional structures over a serial channel”

(PINKER/BLOOM 1990). PINKER and BLOOM describe how specific syntactic structures conform to this language-specific design in ways that provide func- tional advantage: for example, through certain structure-dependent rules (phrase structure) and principles of embedding (recursion) that allow the formulation and expression of infinitely many dis- crete ideas by finite and few means. This enables the multiple thoughts of multiple individuals to be combined, tested in imagination and consequently included or excluded from having a role in some fu- ture action. A population whose individuals could contemplate alternative scenarios in any sequence and at any rate, benefit from the cognitive travails of others, and let conjectures die instead of them- selves, would surely have had an evolutionary ad- vantage over a population that couldn’t.

PINKER and BLOOM offer a well-reasoned and often nonobvious analysis of evolutionary tradeoffs in- volved in opting for linear communication through an auditory medium: visual displays better commu- nicate highly complex topological relationships (“a picture is worth a thousand words”), gestures better convey emotion, and language is inefficient trans- mitting information about taste or smell. For all its insight and plausibility, though, PINKER and BLOOM’s account remains wholly backward-looking: no structural discovery, novel prediction or theoretical breakthrough ensues within the CHOMSKYEAN frame- work that the authors adopt.

According to Jerry FODOR (1998), whose reasoning is close to CHOMSKY’s (1988, p170), it is not likely that natural selection gradually produced an adaptive mutation for a language instinct. Rather, selection pressures unrelated to language simply made human brains a little larger and more complex. Unlike grad- ual and incremental adaptations, a little added com- plexity can go a long way fast to produce multiple novel structures:

“Make the giraffe’s neck just a little bit longer and you correspondingly increase, just by a little, the an- imal’s capacity to reach the fruit at the top of the tree… But make an ape’s brain just a little bigger (or denser, or more folded, or, who knows, grayer) and it’s anybody’s guess what happens to the creature’s behavioural repertoire”. (FODOR 1998).

Much ink has spilled and acrimony vetted in the debate between evolutionary psychologists and the

scientists who criticize them. In the case of language, however, no empirical issue has yet to turn on the debate. Despite very different evolutionary stories, CHOMSKY, FODOR, PINKER and BLOOM fundamentally agree on the specific computational structures that characterize language, on its innateness, on its highly specialized mode of operation (modularity) and on the fact that all of this is uniquely the product of evolution, whether adaptation or by-product.

The Language Module

From a purely logical standpoint, if a mind is able to take fragmentary instances of experience (relative to the richness and complexity of the whole data set) and spontaneously predict (project, generalize) the extension of those scattered cases to an indefinitely large class of intricately related cases, then the infer- ential structure responsible for this prediction can- not possibly derive from the experience. As Hume noted, the structure must be prior to experience, just like the cranes and architects used for construct- ing buildings must exist prior to initial construc- tion: a building does not just build itself. There is no other possibility. What, then, are the physical possi- bilities for a mental structure to be “prior to experi- ence”?

One could hold with SOCRATES or BUDDHA that people are born with a past-life ‘memory’ for the structure of experience they will encounter later in life; or, one could hold with Thomas Aquinas or Kant that God put structure in people’s minds. If, how- ever, one accepts humans as biological creatures whose species attributes emerged through the same evolutionary processes that govern the emergence of all other species, then there appears to be no alterna- tive to a priori mental structures being evolved bio- logical structures (whether as adaptations or by- products of adaptations). To say an evolved biologi- cal structure is ‘innate’ is not to say that every impor- tant aspect of its phenotypic expression is ‘geneti- cally determined’. Biologically poised structures channel development, but do not determine it—like mountains that channel scattered rain into the same mountain–valley river basin (WADDINGTON 1959).

P&P might grant all this, but argue that an all- purpose, domain-general computational device suf- fices to give the mind/brain inferential power be- yond the information given. As an empirical claim, this entails that one should be able to deduce the structural principles and parameters governing any specific domain from general purpose-structures act- ing under the constraints of experience in the rele-

The Case for Modularity: Sin or Salvation?

vant domain. For example, P&P offer that language

“may be closely linked to anterior cingulate and ad- jacent frontal lobe tissues which appear to mediate certain types of pain, feelings of separation distress and thereby social sensitivities” (MACLEAN 1990).

But there is nothing in what we know of general intelligence (e.g., conditioning, association, etc.), pain, distress, social sensitivity, or the structures of anterior cingulate and adjacent frontal lobe tissues, to even remotely hint at the highly structured, and structurally peculiar, principles and parameters of human language. This includes: anaphora (the struc- tural constraints on how even widely separated parts of an expression co-refer), quantification (the struc- tural constraints that ‘who,’ ‘many,’ few,’ etc. im- pose on the syntactic behavior of subjects and ob- jects), negation (it is literally impossible to learn negation through perceived experience or ‘interac- tion,’ as one cannot perceive or interact with some- thing that isn’t), word order (all languages have a specified linear order linking arguments to predicates), and so forth.

Briefly, the central tenet of CHOMSKY’s (2000) ap- proach is that there is a language system, LS, of the human brain. LS reflexively discriminates and cate- gorizes parts of the flux of human experience as ‘lan- guage,’ and develops complex abilities to infer and interpret this highly structured, and structurally pe- culiar, type of human production. In a general sense, there is nothing intrinsically different about LS—in terms of innateness, evolution or universality—than the visual system (VS), immune system (IS), respira- tory system (RS), or any other complex biological system. Learning syntactic structure through ‘social interaction’ is no more plausible an alternative than learning by ‘osmosis’. If a child is initially deprived of linguistic input, the child may never come to ac- quire, or ‘know,’ much in the way of language, just as a kitten initially deprived of normal exposure to the visual world may never come to ‘know’ much by way of object recognition (HUBEL 1988). Still, emo- tion and social interaction no more ‘create’ language in the child’s mind than lighting and object pattern- ing ‘create’ vision in the kitten.

Interacting Modules

LS is no more (or less) ‘autonomous’ from the sur- rounding social environment, or from other mental systems, than VS is detachable from surrounding light and object patterning or from other physical systems (including, in humans, linguistic and other cognitive systems of meaning, MARR 1982). LS and

VS do not exist, and cannot develop, in isolation, but only as subsystems of even more intricate struc- tures. Thus, claims about the biological ‘autonomy’

or ‘modularity’ of LS or VS refer only to a specifiable level of systemic functioning within a system hier- archy. A difficult empirical issue concerns the ex- tent to which other cognitive ‘performance’ systems are themselves specifically adapted for language.

There is little doubt that the sound system is highly structured for access (psycholinguistic experiments with neonates and even fetuses, dichotic listening experiments, comparative phonology, and so forth are compelling). The interface between syntactic and semantic systems is much more obscure. Specu- lation is rampant, debate is furious, and critical ex- periments are few, as might be expected from a relatively new and dynamically changing science.

But progress is being made, slowly and laboriously, as intense research by teams of well-trained investi- gators continues.

Cognitive and developmental psychologists and psycholinguistics have identified a number of struc- tural principles in human cognitive systems that re- late to the interface between LS and these other sys- tems. Among the principles discovered is “the whole–object constraint” (CAREY 1985). Children, whatever their culture or language, apparently as- sume as a default that nouns in general apply to whole objects (e.g., a rabbit) and not to parts of the object (e.g., a piece of a leg, a patch of fur, or spatially separated patches of fur and pieces of leg), or the object–and–its–environment. Children actually have to learn that this is not always the case (just as children have to learn that the shadow of an object is not itself an object, SPELKE et al. 1995). This appar- ently ‘innate bias’ helps to solve the problem of rad- ical indeterminacy of translation. Consider an an- thropologist who visits an exotic tribe for the first time and sees a member of that tribe pointing to something that the anthropologist immediately identifies as a running deer. The anthropologist is fairly safe in assuming that the informant also thought of pointing to a running deer (and not a moving piece of deer, a shifting pattern of deer fur and grass, etc.). This is so even if the informant also believes that the deer is some other person’s ances- tor. If this were not the case—if radical indetermi- nacy were an omnipresent possibility—then anthro- pology would be impossible and ethnography could be only fiction.

Before learning to talk, children first learn to fol- low gaze, engage in joint attention, contingently in- teract with others to achieve goals, and so forth.

Scott Atran

Those who argue that language is acquired through social interaction might be tempted to claim that these facts undermine the modular model of lan- guage (SHANKER 2001). Yet, most of the work in this area—called the child’s ‘theory of mind’ or ‘folkpsy- chology’–focuses on many of the same sorts of ques- tions and approaches that generative grammarians do: How is the child able to reliably infer such rich mental structures about other minds from a few ges- tures and without mastery of language? And how is it that children also infer that people’s mental struc- tures (intentions) can cause others to act a distance (without any physical contact)? The emerging con- sensus in the field is that children are biologically endowed with a ‘theory of mind’ (ToM) that matures in predictable ways over the first three years of life (LESLIE 1994; BARON-COHEN 1995; SPELKE et al 1995).

Some features of developing ToM interface with ma- turing features of developing LS in systematic ways that are only now being experimentally studied. This is one of the most exciting and productive research areas in cognitive and developmental psychology (to judge from journal publications and grant funding), and also one of the most thoroughly ‘modular’.

Interactions between modular cognitive systems are complex and difficult to tease apart, perhaps more so than interactions among various bodily sys- tems. For example, belief in supernatural agents, which characterizes all religions in all societies, in- volves a host of modular expectations and interfac- ing: folkmechanics, folkpsychology, folkbiology, primary and secondary emotions, predator–prey schema, and so forth (BOYER 1994; BARRETT/NYHOFF

2001; ATRAN in press). Nevertheless, predictive theo- ries are being proposed and empirically tested.

Are Modules Adaptations or

Evolutionary By-Products of Adaptations?

Leading evolutionary biologists, such as Stephen GOULD and Richard LEWONTIN (1979) describe the products of higher-order human cognition as evolu- tionary ‘by-products’ or ‘spandrels’. This is a stance P&P adopt. A spandrel is an architectural term for the structural form or space that arises as a necessary concomitant to another decision in design, and is not designed to have any direct utility in itself. For example, the space beneath a flight of stairs is a by- product of constructing an inclined stairway rather than a vertical ladder. The fact that people might subsequently use this ‘leftover’ space for storage does not entail that the space was designed to be a storage space.

For GOULD and LEWONTIN, higher-order human cognitive structures originated as functionless span- drels that have been subsequently modified under cultural selection rather than natural selection. Bio- logically functionless, or nearly functionless, span- drels supposedly include: religion, writing, art, sci- ence, commerce, war and play. These evolutionary by-products are cultural ‘mountains’ to the biologi- cally ‘adaptive molehill’ (GOULD 1991, pp58–59;

WILLIAMS 1992, pp77–79). On this account, evolu- tionary psychology would have little to reveal about the emergence and structure of such culturally-elab- orated spandrels: “The number and complexity of these spandrels should increase with the intricacy of the organism under consideration. In some region within a spectrum of rising complexity, the number and importance of useable and significant spandrels will probably exceed the evolutionary import of the primary adaptation” (GOULD 1997, pp10754–10755;

cf. FODOR 1998).

GOULD’s account of the emergence of distinctly human cognitions as spandrels of a big brain is hardly convincing. The very notion of a ‘big’ or

‘large’ or ‘complex’ brain is too vague to empirically constrain the evolutionary story about how it might have evolved. The big brain is taken as an adaptation from which all cognitive spandrels arise. But an ad- aptation to what? A design for what? What are its evolutionary-relevant computational structures?

Big brain stories tend to be even broader in scope and handwaving than language–evolution stories. Hu- man brains supposedly broke away from ape brains under selection pressures that run the gamut from runaway social competition (ALEXANDER 1989), to gut reduction (AIELLO/WHEELER 1995), to hunting large game (HILL 1999), to niche construction (LA- LAND et al. 2000), to runaway sexual selection (MILLER 2000), and so on. Alternatively, the big brain primarily evolved as a conduit for culture (HARRIS

1975), as a vehicle for language (JERISON 1976), or as a host for the independently evolving ideas, or

‘memes,’ that compete to colonize it (DENNETT 1995;

BLACKMORE 1999). Finally, the big brain may have evolved under any number ‘positive feedback’ path- ways involving some or all of the factors mentioned.

But even if true, and even if we knew the reasons (selection pressures responsible for) why, we are un- likely to learn anything of particular interest from all this about how the mind works. The notion of a big brain is as uninformative about cognitive structures and functions cognition as the notion of big body is about bodily structures and functions. Never mind exaptations, such as chins, or spandrels, such as

The Case for Modularity: Sin or Salvation?

palm lines. From the fact of a bigger body, what could one possibly deduce about hearts, livers, kid- neys, hands, faces, placentas and so on? And never mind exaptations, such as language, or spandrels, such as religion. From the fact of a bigger (or denser, or more folded, or grayer) brain, what could one pos- sibly deduce about perception, emotion, categoriza- tion, inference or any of the other capabilities hu- mans share with apes but in more vastly elaborated form? Probably nothing at all.

It may well be true that little insight is to be gained into higher-order human cognitive structures by considering possible evolutionary origins and func- tions. Adopting the hypothesis of GOULD and LEWONTIN or P&P, which assumes this truth, is prac- tically guaranteed to block insight, whether or not it is true. By contrast, adopting evolutionary psychol- ogy’s requirement that candidates for exaptations and spandrels be described, as far as possible, with reference to evolved adaptations, then it might be possible to find out if the hypothesis is true or not.

If it is true, then evolutionary psychology would have provided the empirical evidence that shows it to be a significant and surprising scientific insight, and not one that depends entirely on intuition, anal- ogy, eloquence or wishful thinking. If it is not true, then evolutionary psychology will have helped to discover something new about human nature.

Three Examples of Modularity: Naïve Mechanics, Theory of Mind, Folkbiology

Ever since CHOMSKY jump-started the ‘cognitive rev- olution,’ successors to the behaviorists who believed in an all-powerful general thinking device have tried to reconcile CHOMSKY’s insights with faith in flexible intelligence by reluctantly granting some specificity to language, and language alone.

But cognitive psychology today concentrates more on discovery and exploration of domain-specific mechanisms than on general-purpose computation.

Each such device has a particular ‘content-bias’ in that it targets some particular domain of stimuli in the world (‘set of inputs’): for example, the edges and trajectories of rigid three-dimensional bodies that move by physical contact between them (mechanics), the contingent motion a self-pro- pelled actors that can coordinate interactions with- out having physical contact (agency), or the behaviors and appearances of nonhuman living creatures (species relations). The particular inferen- tial structure of each domain-specific processor then takes the isolated exemplars (or relatively poor

samples) of the stimulus-set actually encountered in a person’s life, and spontaneously projects these rel- atively fragmentary instances onto richly-struc- tured categories (‘classes of output’) of general relevance to our species: for example, the objects and kinds of folkphysics (naïve mechanics), folkpsychology (ToM) and folkbiology. Much work on domain-specificity has developed, and now develops, independently of sociobiology or evolu- tionary psychology (ATRAN 1989; HIRSCHFELD 1996;

KEIL 1989; LESLIE 1994; CAREY/SPELKE 1994; SPERBER

1985).

Within the emerging paradigm of cognitive do- main-specificity, there is much speculation and con- troversy—again, as might be expected in any newly emerging science. For example, there are competing accounts of how human beings acquire basic knowl- edge of the everyday biological world, including the categorical limits of the biological domain and the causal nature of its fundamental constituents. One influential view of conceptual development in folk- biology has been articulated by Susan CAREY and her collaborators (CAREY 1985; CAREY/SPELKE 1994;

JOHNSON/CAREY 1998). On this view, young chil- dren’s understanding of living things is embedded in a folkpsychological, rather than folkbiological, explanatory framework. Only by age 7 do children begin to elaborate a specifically biological frame- work of the living world, and only by age 10 does an autonomous theory of biological causality emerge that is not based on children’s understanding of how humans think and behave.A competing view is that folkbiology and folkpsychology emerge early in childhood as largely independent domains of cogni- tion that are clearly evident by ages 4 or 5, and which may be innately differentiated (ATRAN 1987; KEIL

1989; GELMAN/WELLMAN 1991; HATANO/INAGAKI

1999).

To address this issue, my colleagues and I carried out a series of cross-cultural experiments (LÓPEZ et al 1997; MEDIN et al. 1997; ATRAN et al. 2001). One set of experiments shows that by the age of 4–5 years (the earliest age tested in this regard) urban Ameri- can and Yukatek Maya children employ a concept of innate species potential, or underlying essence, as an inferential framework for understanding the affilia- tion of an organism to a biological species, and for projecting known and unknown biological proper- ties to organisms in the face of uncertainty. Another set of experiments shows that the youngest Maya children do not have an anthropocentric under- standing of the biological world. Children do not initially need to reason about nonhuman living

Scott Atran

kinds by analogy to human kinds. The fact that American children show anthropocentric bias ap- pears to owe more to a difference in cultural expo- sure to nonhuman biological kinds than to a basic causal understanding of folkbiology per se. Together, the first two sets of experiments suggest that folkpsy- chology can’t be the initial source of folkbiology.

They also indicate that to master biological science, people must learn to inhibit activation of universal dispositions to view species essentialistically and to see humans as inherently different from other ani- mals (ATRAN 1990, 1998).

A third set of results show that the same taxo- nomic rank is cognitively preferred for biological in- duction in two diverse populations: people raised in the Midwestern USA and Itza’ Maya of the Lowland Mesoamerican rainforest (ATRAN et al. 1997; COLEY

et al. 1997). This is the generic species—the level of oak and robin. These findings cannot be explained by domain-general models of similarity because such models cannot account for why both cultures prefer species-like groups in making inferences about the biological world, although Americans have rela- tively little actual knowledge or experience at this level. In fact, general relations of perceptual similar- ity and expectations derived from experience pro- duce a ‘basic level’ of recognition and recall for many Americans that corresponds to the superordinate life-form level of folkbiological taxonomy—the level of tree and bird (ROSCH et al. 1976). Still, Americans prefer generic species for making inductions about the distribution of biological properties among or- ganisms, and for predicting the nature of the biolog- ical world in the face of uncertainty. This supports the idea of the generic-species level as a partitioning of the ontological domains of plant and animal into mutually exclusive essences that are assumed (but not necessarily known) to have unique underlying causal natures. The implication from these experi- ments is that folkbiology may well represent an evo- lutionary design: universal taxonomic structures, centered on essence-based generic species, are argu- ably routine products of our ‘habits of mind’, which may be in part naturally selected to grasp relevant and recurrent ‘habits of the world’.

Evidence for Modularity

The evolutionary argument for a naturally-selected cognitive disposition, such as folkbiology, involves converging evidence from a number of venues:

Functional design (analogy), ethology (homology), universality, precocity of acquisition, independence

from perceptual experience (poverty of stimulus), selective pathology (cerebral impairment), resis- tance to inhibition (hyperactivity), and cultural transmission. None of these criteria may be neces- sary, but the presence of all or some is compelling, if not conclusive.

1. Functional Design. All organisms must function to procure energy to survive, and they also must pro- cure (genetic) information for recombination and reproduction (ELDREDGE 1986). The first require- ment is primarily satisfied by other species, and an indiscriminate use of any individual of the other spe- cies (e.g., energy-wise, it does not generally matter which chicken or apple you eat). The second require- ment is usually only satisfied by genetic information unique to individual conspecifics (e.g., genetically, it matters who is chosen as a mate and who is con- sidered kin). On the one hand, humans recognize other humans by individuating them with the aid of species-specific triggering algorithms that ‘automat- ically’ coordinate perceptual cues (e.g., facial recog- nition schemata, gaze) with conceptual assumptions (e.g., intentions) (BARON-COHEN 1995). On the other hand, people do not spontaneously individuate the members of other species in this way, but as exam- plars of the (generic) species that identifies them as causally belonging to only one essential kind.

Natural selection basically accounts only for the appearance of complexly well-structured biological traits that are designed to perform important func- tional tasks of adaptive benefit to organisms. In gen- eral, naturally selected adaptations are structures functionally “perfected for any given habit” (DAR- WIN 1883, p140), having “very much the appearance of design by an intelligent designer… on which the wellbeing and very existence of the organism de- pends” (WALLACE 1901, p138). Plausibly, the univer- sal appreciation of generic species as the causal foun- dation for the taxonomic arrangement of biodiversity, and for taxonomic inference about the distribution of causally-related properties that un- derlie biodiversity, is one such functional evolution- ary adaptation. But a good story is not enough.¹

2. Ethology. One hallmark of adaptation is a phylo- genetic history that extends beyond the species in which the adaptation is perfected: for example, ducklings crouching in the presence of hawks, but not other kinds of birds, suggests dedicated mecha- nisms for something like species recognition. Some nonhuman species can clearly distinguish several different animal or plant species (CERELLA 1979;

The Case for Modularity: Sin or Salvation?

LORENZ 1966; HERRNSTEIN 1984). Vervet monkeys even have distinct alarm calls for different predator species or groups of species: snake, leopard and chee- tah, hawk eagle, and so forth (HAUSER 2000). Chim- panzees may have rudimentary hierarchical group- ings of biological groups within groups (BROWN/ BOYSEN in press). To be sure, the world itself is nei- ther chaos nor flux: species are often locally self- structuring entities that are reproductively and eco- logically isolated from other species through natural selection. But there is no a priori reason for the mind to always focus on categorizing and relating species qua species, unless doing so served some adaptive function (e.g., it makes little difference which tiger could eat a person or which mango a person could eat). And the adaptive functions of organisms rarely, if ever, evolve or operate in nature as all-purpose mechanisms.

3. Universality. Ever since the pioneering work of BERLIN and his colleagues, evidence from ethnobiol- ogy and experimental psychology has been accumu- lating that all human societies have similar folkbio- logical structures (BERLIN et al. 1973; BERLIN 1992;

ATRAN 1990, 1999). These striking cross-cultural sim- ilarities suggest that a small number of organizing principles universally define folkbiological systems.

Basic aspects of folkbiological structure (e.g., taxo- nomic ranking, primacy of generic-species) seem to vary little across cultures as a function of theories or belief systems.

4. Ease of Acquisition. Acquisition studies indicate a precocious emergence of essentialist folkbiological principles in early childhood that are not applied to other domains (GELMAN/WELLMAN 1991; KEIL 1994;

HATANO/INAGAKI 1999; ATRAN et al. 2001).

5. Independence from Perceptual Experience.

Experiments on inferential processing show that that humans do not make biological inductions pri- marily on the basis of perceptual experience or any general similarity-based metric, but on the basis of imperceptible causal expectations of a peculiar, es- sentialist nature (ATRAN et al 1997; COLEY et al 1997).

6. Pathology. Cerebral impairments (WILLIAMS syn- drome, brain lesions caused by certain types of her- pes virus, etc.) suggest selective retention or loss of folkbiological taxonomies or of particular taxo- nomic ranks. Neuropsychological studies have re- ported a pathological performance in recognition at the life-form and generic-species levels (e.g., recog-

nizing an item as an animal but not as a bird or robin), and dissociation at the life-form level (e.g., not recognizing items as trees). Existing studies, however, do not say anything about the generic-spe- cies rank as the preferred level of representation for reasoning, perhaps because of methodology (linked to averaging over items and failure to include sets of generic species) (WARRINGTON/SHALLICE 1984; SAR- TORI/JOB 1988; JOB/SURIAN 1998).

7. Inhibition and Hyperactivity. One characteris- tic of an evolved cognitive disposition is evident dif- ficulty in inhibiting its operation (HAUSER 2000).

Consider beliefs in biological essences. Such beliefs greatly help people explore the world by prodding them to look for regularities and to seek explana- tions of variation in terms of underlying patterns.

This strategy may help bring order to ordinary cir- cumstances, including those relevant to human sur- vival. But in other circumstances, such as wanting to know what is correct or true for the cosmos at large, such intuitively ingrained concepts and beliefs may hinder more than help. For example, the essentialist bias to understand variation in terms of deviance is undoubtedly a hindrance to evolutionary thinking.

In some everyday matters, the tendency to essential- ize or explain variation in terms of deviation from some essential ideal or norm (e.g., people as mental or biological ‘deviants’) can be an effortlessly ‘natu- ral’ but wrong way to think.

Because intuitive notions come to us so naturally they may be difficult to unlearn and transcend. Even students and philosophers of biology often find it difficult to abandon commonsense notions of species as classes, essences or natural kinds in favor of the concept of species as a logical individual—a genealog- ical branch whose endpoints are somewhat arbi- trarily defined in the phyletic tree and whose status does not differ in principle for that of other smaller (variety) and larger (genus) branches. Similarly, rac- ism—the projection of biological essences onto social groups—seems to be a cognitively facile and cultur- ally universal tendency (HIRSCHFELD 1996). Although science teaches that race is biologically incoherent, racial or ethnic essentialism is as notoriously difficult to suppress as it is easy to incite (GIL-WHITE 2001).

8. Cultural Transmission. Human cultures favor a rapid selection and stable distribution of those ideas that: 1) readily help to solve relevant and recurrent environmental problems, 2) are easily memorized and processed by the human brain, and 3) facilitate the retention and understanding of ideas that are

Scott Atran

more variable (e.g ., religion) or difficult to learn (e.g., science) but contingently useful or important.

Folkbiological taxonomy readily aids humans every- where in orienting themselves and surviving in the natural world. Its content tends to be stable within cultures (high interinformant agreement, substan- tial historical continuity) and its structure isomor- phic across cultures (BERLIN et al. 1973; LÓPEZ et al.

1997). Folkbiological taxonomy also serves as a prin- cipled basis for transmission and acquisition of more variable and extended forms of cultural knowledge, such as certain forms of religious and scientific belief (ATRAN 1990, 1998).

Consider, for example, the spontaneous emer- gence of totemism—the correspondence of social groups with generic species—at different times and in different parts of the world. Why, as LÉVI-STRAUSS

(1963) aptly noted, are totems so “good to think”?

In part, totemism uses representations of generic species to represent groups of people; however, this pervasive metarepresentational inclination argu- ably owes its recurrence to its ability to ride piggy- back on folkbiological taxonomy. Generic species and groups of generic species are inherently well- structured, attention-arresting, memorable and readily transmissible across minds. As a result, they readily provide effective pegs on which to attach knowledge and behavior of less intrinsically well- determined social groups. In this way totemic groups can also become memorable, attention-arresting and transmissible across minds. These are the condi- tions for any idea to become culturally viable (see SPERBER 1996 for a general view of culture along the lines of an ‘epidemiology of representations’). A sig- nificant feature of totemism that enhances both memorability and its capacity to grab attention is that it violates the general behavior of biological spe- cies: members of a totem, unlike members of a ge- neric species, generally do not interbreed, but only mate with members of other totems in order to cre- ate a system of social exchange. Notice that this vio- lation of core knowledge is far from arbitrary. In fact, it is such a pointed violation of human beings’ intu- itive ontology that it readily

mobilizes most of the assump- tions people ordinarily make about biology in order to bet- ter help build societies around the world (ATRAN/SPERBER

1991).

In brief, modularized struc- tures—such as those which produce folkmechanical, folk-

psychological and folkbiological concepts—are spe- cial players in cultural evolution. Their native stabil- ity derivatively attaches to more variable and difficult–to–learn representational forms, thus en- hancing the latter’s prospects for regularity and re- currence in transmission within and across cultures.

This includes knowledge that cumulatively enriches (e.g., to produce folk expertise), overrides (e.g., to produce religious belief) or otherwise transcends (e.g., to produce science) the everyday ontology pre- scribed by our evolved cognitive modules.

Conclusion: Neither Sin nor Salvation, but a Promissory Note.

Despite the initial independence of work in domain- specificity, there is now increasing convergence in the ways cognitive anthropologists and psycholo- gists, and evolutionary biologists and psychologists, think about related issues. The general consensus is that domain-specific mechanisms likely evolved over millions of years of biological and cognitive evolution to deal with specific sorts of relevant and recurrent problems in ancestral environments (‘task demands’), such as recognizing inert objects (e.g., rocks), reducing biodiversity to causally manageable proportions (e.g., species), or anticipating agents (e.g., the intentions of potential predators or prey).

In the computer lingo that now dominates the field, the ‘computational mind’ consists of a variety of dis- tinct, task-specific information-processing devices.

Nothing from work on subcortical emotions or neo- cortical principles of general association has pro- vided, or hints at providing, the slightest insight into the cognitive structures and processes discov- ered so far in these emerging fields.

Evolutionary psychology is still in its infancy and is not yet the new scientific paradigm some would love and others hate. Some find the idea of

‘Stone Age Minds for a Space Age World’ bold and irreverent, many find it false and demeaning, oth- ers find it ridiculous. As the field stands now, all may garner uncertain support for their position.

This new field surely will not solve all of the problems its fervent supporters say it will.

But neither will it face the massive road blocks to under- standing that its unrelenting detractors see at every turn. At present, the field is a promis- sory note, much as DARWIN’s theory was at its beginning.

Scott Atran, CNRS–Institut Jean Nicod, 1 bis, avenue de Lowendal, F-75007 Paris, France, and School of Natural Resources and Environment, Dana Building, 430 East Uni- versity, Ann Arbor, Michigan 48109-1115, USA. Email: satran@umich.edu

Author’s address

The Case for Modularity: Sin or Salvation?

Little by little, biologists were able to deliver on DARWIN’s promises. This process has speeded up con- siderably—almost irreversibly—thanks to access at the molecular level. There is still a long way to go.

Through recent advances in cognitive science, evo- lutionary psychology has gained entrance to mental structure, and so potentially to the brain’s evolved neural architecture. It has an even longer way ahead:

much less is currently known about how the mind/

brain works than how body cells function. Perhaps, in the end, evolutionary psychology’s interpreta- tions of complex mental designs as telltale signs of ancient environments will prove no truer than phre- nology’s readings of bumps and other conforma- tions of the skull as indications of mental faculties and character (phrenology was a very serious and hotly debated discipline a century ago). Then again perhaps not, which makes the effort worthwhile.

Note

1 Although the adaptive relationship of structure to function is often manifest, as with the giraffe’s neck or the rhinocer- us’s horns, often it is not. In such cases, evolutionary the- orists adopt a strategy of ‘reverse engineering’. Reverse engineering is what military analysts do when a weapon from an enemy or competitor in the arms market falls into their hands and they try to figure out exactly how it was put together and what it can do. Reverse engineering is easiest, of course, if the structure contains some signature of its function, like trying to figure out what a toaster does given the telltale sign of toasted bread crumbs left inside.

But in many cases recognizing the appropriate signs already requires some prior notion of what function the structure may have served. Thus, after a century and a half of debate,

it is only now that scientists clearly favor the hypothesis that bipedality was primarily selected to enhance field of view. Comparative studies of humans with bipedal birds and dinosaurs, as well as experiments comparing energy expenditure and running speed in two-footed versus four- footed running and walking, appear to exclude the compet- ing hypotheses that bipedality evolved for running or en- ergy conservation. For most higher-order human cognitive faculties, however, there may be little useful comparative evidence from elsewhere in the animal kingdom. This is because of their apparent structural novelty, poor represen- tation in the fossil record (e.g., stone tools tell little of lan- guae or theory of mind) and lack of surviving intermediate forms. The moral is that reverse engineering can be helpful, and occasionally successful, but success is by no means guaranteed even in the richest of evidentiary contexts.

References

Aiello, L./Wheeler, P. (1995) The expensive-tissue hypothe- sis. Current Anthropology 36:199–221.

Alexander, R. (1989) Evolution of the human psyche. In:

Stringer, C. (ed) The human revolution. The University of Edinburgh Press: Edinburgh.

Atran, S. (1987) Ordinary constraints on the semantics of liv- ing kinds: A commonsense alternative to recent treat- ments of natural-object terms. Mind and Language 2: 27–

63.

Atran, S. (1989) Basic conceptual domains. Mind and Lan- guage 4: 7–16.

Atran, S. (1990) Cognitive foundations of natural history: To- wards an anthropology of science. Cambridge University Press: Cambridge.

Atran, S. (1998) Folkbiology and the anthropology of sci- ence: Cognitive universals and cultural particulars. Behav- ioral and Brain Sciences 21: 547–609.

Atran, S. (1999) Itzaj Maya folk-biological taxonomy. In: Me- din, D./Atran, S. (eds) Folk biology. MIT Press: Cambridge MA.

Atran, S./Estin, P./Coley, J./Medin, D. (1997) Generic species and basic levels: Essence and appearance in folk biology.

Journal of Ethnobiology 17: 22–45.

Atran, S./Medin, D./Lynch, E./Vapnarsky, V./Ucan Ek’, E./

Sousa, P. (2001) Folkbiology doesn’t come from folkpsy- chology: Evidence from Yukatek Maya in cross-cultural perspective. Journal of Cognition and Culture 1: 3–43.

Atran, S. (in press) In gods we trust: The evolutionary land- scape of religion. Oxford University Press: New York.

Atran, S./Sperber, D. (1991) Learning without teaching: Its place in culture. In: Tolchinsky-Landsmann, L. (ed) Cul-

ture, schooling and psychological development. Ablex:

Norwood NJ.

Barrett, J./Nyhof, M. (2001) Spreading non-natural concepts:

The role of intuitive conceptual structures in memory and transmission of cultural materials. Journal of Cognition and Culture 1: 69–100.

Baron-Cohen, S. (1995) Mindblindness. The MIT Press: Cam- bridge MA.

Berlin, B./Breedlove, D.,/Raven, P. (1973) General principles of classification and nomenclature in folk biology. Ameri- can Anthropologist 74: 214–242.

Berlin, B. (1992) Ethnobiological classification. Princeton University: Princeton.

Blackmore, S. (1999) The meme machine. Oxford University Press: Oxford.

Boyer, P. (1994) The naturalness of religious ideas. University of California Press: Berkeley.

Brown, D./Boysen, S. (in press) Spontaneous discrimination of natural stimuli by chimpanzees (Pan troglodytes). Jour- nal of Comparative Psychology.

Buss, D. (1999) Evolutionary psychology. Allyn & Bacon:

Boston.

Carey, S. (1985) Constraints on semantic development. In:

Mehler, J./Fox, R. (eds) Neonate cognition. Erlbaum: Hills- dale NJ.

Carey, S./Spelke, E. (1994) Domain-specific knowledge and conceptual change. In: Hirschfeld, L./Gelman, S. (eds) Mapping the mind. Cambridge University Press: New York.

Cerella, J. (1979) Visual classes and natural categories in the pigeon. Journal of Experimental Psychology Human per- ception and performance 5: 68–77.

Chomsky, N. (1982) Lectures on government and binding.

Scott Atran

Foris Publications: Dordrecht. ***NOT_MENTIONED***

Chomsky, N. (1988) Language and problems of knowledge.

MIT Press: Cambridge MA.

Chomsky, N. (2000) Minimalist inquiries: The framework. In:

Martin, R./Michaels, D./Uriagereka, J. (eds) Step by step.

MIT Press: Cambridge MA.

Coley, J./Medin, D./Atran, S. (1997) Does rank have its privi- lege? Inductive inferences in folkbiological taxonomies.

Cognition 63: 73–112.

Damasio, A. (1994) Descartes’ error: Emotion, reason, and the human brain. Avon Books: New York.

Darwin, C. (1883) On the origins of species by means of natu- ral selection. 6th edition. Appleton: New York. Originally published 1872.

Dennett, D. (1995) Darwin’s dangerous idea: Evolution and the meaning of life. Simon and Schuster: New York.

Eldredge, N. (1986) Information, economics, and evolution.

Annual Review of Ecology and Systematics 17: 351–369.

Fodor, J. (1998) The trouble with psychological Darwinism.

London Review of Books 20(2): ***PAGE_NUMBERS***

Gelman, S./Wellman, H. (1991) Insides and essences. Cogni- tion 38: 214–244.

Gil-White, F. (2001) Are ethnic groups biological “species” to the brain? Current Anthropology 42(4):

***PAGE_NUMBERS_MISSING***

Gould, S. (1991) Exaptation: A crucial tool for evolutionary psychology. Journal of Social Issues 47: 43–46.

Gould, S. (1997) The exaptive excellence of spandrels as a term and prototype. Proceedings of the National Academy of Sciences USA 94: 10750–10755.

Gould, S./Lewontin, R. (1979) the spandrels of San marco and the Panglossian paradigm. Proceedings of the Royal Society of London B. 205: 581–598.

Griffiths, P. (1997) What emotions really are. University of Chicago Press: Chicago.

Harris, M. (1975) Culture, people, nature: An introduction to general anthropology. 2nd edition. Thomas Crowell: New York.

Hatano, G./Inagaki, K. (1999) A developmental perspective on informal biology. In: Medin, D./Atran, S. (eds) Folk bi- ology. MIT Press: Cambridge MA.

Hauser, M. (2000) What animals really think. Henry Holt &

Company: New York.

Herrnstein, R. (1984) Objects, categories, and discriminative stimuli. In: Roitblat, H. (ed) Animal cognition. Erlbaum:

Hillsdale NJ.

Hill, K. (1999) Evolution of the human life course. Paper pre- sented to The Evolution and Human Adaptation Program, Institute for Social Research. The University of Michigan, Ann Arbor, February 9, 1999.

Hirschfeld, L. (1996) Race in the making. MIT Press: Cam- bridge MA.

Hubel, D. (1988) Eye, brain, and vision. Scientific American:

New York.

Jerison, H. (1976) The paleoneurology of language. In: Har- nad, S./Steklis, H./Lancaster, J. (eds.) Origins and evolution of language and speech. Annals of the New York Academy of Sciences 280: 370–382.

Job, R./Surian, L. (1998) A neurocognitive mechanism for folk biology? Behavioral and Brain Sciences 21: 577–578.

Johnson, S./Carey, S. (1998) Knowledge enrichment and conceptual change in folkbiology: Evidence from people with Williams Syndrome. Cognitive Psychology 37: 156–

200.

Keil, F. (1989) Concepts, kinds, and cognitive development.

MIT Press: Cambridge MA.

Keil, F. (1994) The birth and nuturance of concepts by do- mains. In: Hirschfeld, L./Gelman, S. (eds.) Mapping the mind. Cambridge University Press: New York.

LeDoux, J. (1996) The emotional brain. Simon and Schuster:

New York.

López, A./Atran, S./Coley, J./Medin, D.,/Smith, E. (1997) The tree of life: Universals of folk-biological taxonomies and inductions. Cognitive Psychology 32: 251–295.

Laland, K./Olding-Smee, J./Feldman, M. (2000) Niche con- struction, biological evolution and cultural change. Behav- ioral and Brain Sciences 23: 131–146.

Leslie, A. (1994) ToMM, ToBy, and agency. In: Hirschfeld, L./

Gelman, S. (eds) Mapping the mind: Domain-specificity in cognition and culture. Cambridge University Press: New York.

Lévi-Strauss, C. (1963) The bear and the barber. The Journal of the Royal Anthropological Institute 93: 1–11.

Lorenz, K. (1966) The role of gestalt perception in animal and human behavior. In: White, L. (ed) Aspects of form. Indi- ana University Press: Bloomington.

MacLean, P. (1990) The triurne brain in evolution. Plenum Press: New York.

Marr, D. (1982) Vision. W. H. Freeman: New York.

Medin, D./Lynch, E./Coley, J./Atran, S. (1997) Categoriza- tion and reasoning among tree experts: Do all roads lead to Rome? Cognitive Psychology 32: 49–96.

Miller, G. (2000) The mating mind: How sexual choice shaped the evolution of human nature. Doubleday Books:

New York.

Nesse, R./Lloyd, A. (1992) The evolution of psychodynamic mechanisms. In. Barkow, J./Cosmides, L./Tooby, J. (eds) The adapted mind. Oxford University Press: New York.

Panksepp, J./Panksepp, J.B. (2001) The seven sins of evolu- tionary psychology. Evolution and Cognition 6(2): 108–

131.

Pinker, S./Bloom, P. (1990) Natural language and natural se- lection. Behavioral and Brain Sciences 13: 707–727.

Rosch, E./Mervis, C./Grey, W./Johnson, D.,/Boyes-Braem, P.

(1976) Basic objects in natural categories. Cognitive Psy- chology 8: 382–439.

Sartori, G./Job, R. (1988) The oyster with four legs: A neuro- psychological study on the interaction of semantic and vi- sual information. Cognitive Neuropsychology 5: 105–132.

Shanker, S. (2001) What a child knows when she knows what a name is: The non-Cartesian view of language acquisition.

Current Anthropology 42(4):

***PAGE_NUMBERS_MISSING***

Spelke, E./Phillips, A./Woodward, A. (1995) Infants’ knowl- edge of object motion and human action. In: Sperber, D./

Premack, D./Premack, A. (eds) Causal cognition: A multi- disciplinary debate. Clarendon Press: Oxford.

Sperber, D. (1985) Anthropology and psychology: Towards an epidemiology of representations. Man 20: 73–89.

Sperber, D. (1996) Explaining culture: A naturalistic ap- proach. Blackwell: Oxford.

Thornhill, R./Palmer, C. (2000) A natural history of rape. MIT Press: Cambridge MA.

Waddington, C. (1959) Canalisation of development and the inheritance of acquired characteristics. Nature 183: 1654–

1655.

Wallace, A. (1901) Darwinism. Macmillan: New York. Origi- nally published 1889.

Warrington, E./Shallice, T. (1984) Category specific impair- ments. Brain 107: 829–854.

Williams, G. (1992) Natural selection: Domains, levels and challenges. Oxford University Press: New York.